The proposed research explores the role of the human hippocampus in implicit context learning using functional magnetic resonance imaging. While it has long been known that the hippocampus is critically involved in context learning, existing studies have largely been limited to explicit, or consciously known contextual relations. However, context learning can occur without explicit awareness. Only recently have implicit context task begun to be systematically explored, and in each case the hippocampus is implicated (e.g.,Chun & Phelps, 1999; Graham et al., 2006; Greene et al., 2007; Ryan, Althoff, Whitlow, & Cohen, 2000). However, the neural mechanisms of implicit context learning have only begun to be explored. The role of implicit context has recently become an area of interest for drug abuse and addiction researchers. During recovery, implicit contextual cues previously learned to predict drug delivery serve as a trigger for craving and relapse. The existence of implicit context cues may make recovery more difficult because the cues are not readily identified as triggers and are not extinguished. Presently, little is known about implicit context learning and its modification, particularly the role of the hippocampus. The proposed research has two aims relevant to understanding implicit context as it relates to the treatment of drug addiction. First it explores the neural mechanisms for acquisition and retrieval of implicit context as a predictor. Second, it examines the neural systems involved when the predictive value of the context changes. We will use the Contextual Cueing task (Chun & Jiang, 1998; Greene et al., 2007), an established hippocampal-dependent, contextual learning paradigm which can be performed either implicitly or explicitly. The task is to locate a target (a rotated "T") among a field of distractors (each a rotated "L"). While half of all arrays are novel, twelve arrays repeat throughout. In the repeated arrays the pattern of distractors predicts the target location. Without recognizing repeated arrays, participants become faster at locating targets in repeated arrays. The explicit version is accomplished by informing participants at the outset that they can learn to use some pattern as a predictor for the target location (Chun & Jiang, 2003; Greene & Gross, 2003). We will also manipulate the predictive value of the repeated arrays. Here, in the initial training phase for repeated arrays the context will predict target location. In the second training phase the repeated arrays will have no predictive value (the target location is randomized). And in the third training phase, the predictive value of the repeated arrays is restored. Behavioral and hemodynamic responses will be evaluated. Hemodynamic responses will use whole brain analysis with an emphasis on the hippocampus and other medial temporal lobe structures. The proposed research described here has direct implications for understanding basic neural mechanisms of context learning, both implicit and explicit in humans. During drug rehabilitation, implicit context cues may trigger craving resulting in relapse. Implicit context is a relatively new research area that has not been adequately explored and its mechanisms are not understood. This work may provide critical insights for developing more effective treatments for preventing eliminating the triggers that induce craving and lead to relapse. [unreadable] [unreadable] [unreadable]